OpenAI Brings Its Models and Codex to Google Cloud, Completing a Sweep of All Three US Hyperscalers

OpenAI announced on Tuesday that its frontier models and Codex agent workflows are coming to Google Cloud, with availability in the Model Garden rolling out over the coming weeks. Enterprise customers will also be able to buy the models through Google Cloud Marketplace and pay using their existing Google Cloud spend commitments — a detail that matters for the large fraction of enterprises already locked into multi-year Google contracts.

The integration does not stop at inference endpoints. OpenAI said its models will also surface in Google's Antigravity coding tool and in Gemini Agent, giving developers a path to call GPT models from inside Google's own agent and developer environments. Both companies framed the deal as combining Google Cloud's access controls, data governance tooling and infrastructure with OpenAI's model stack for building enterprise agent workflows.

Strategically, the announcement completes a notable sweep: OpenAI's models are now or will soon be available across all three major US hyperscalers — Microsoft Azure, Amazon Web Services and Google Cloud. OpenAI was careful to note that Microsoft remains its primary cloud partner, but the distribution footprint keeps widening beyond it. Google joined Amazon earlier this month in shipping developer tools that make it easier for PyTorch users to move workloads off NVIDIA GPUs, and hosting a rival lab's models is another step in the same direction.

For Google, the deal is a quietly pragmatic one. Model Garden has been filling with third-party frontier models, and renting inference for OpenAI's models converts a competitive threat into metered revenue — paid, in many cases, with committed Google Cloud spend that enterprises would spend anyway.

For OpenAI, every additional cloud channel reduces dependence on any single infrastructure partner ahead of an expected public listing, and meets enterprise buyers where their budgets and compliance reviews already live. The company's enterprise push is accelerating on other fronts too: ChatGPT's advertising business recently expanded to 26 additional countries, bringing the total to 83 markets with more than 100,000 advertisers.

The competitive picture is now genuinely three-sided. Microsoft gets OpenAI's loyalty in name, AWS and Google Cloud get OpenAI's weights on their metal, and enterprises get to negotiate. What none of the three clouds get any longer is exclusivity — and pricing power in enterprise AI inference is likely to erode accordingly.